Oct 2024 – Feb 2026 · 1 yr 4 mos · Remote

• Fraud detection (Series-C fintech): Ensemble ML pipeline preventing $1.85M/yr in chargeback losses. 92% recall, 88% precision @ 2% FPR, P95 <100ms, ~300 TPS.
• RAG optimization (seed-stage AI platform): Migrated from Pinecone to pgvector + HNSW, cutting hosting cost 80%. Built hybrid retrieval powering 30K daily queries at 84% recall.

Apr 2024 – Aug 2024 · 4 mos · San Francisco, CA

• Built an AutoML orchestration layer that cut model iteration cycles from 3 weeks to 5 days across 15 POC customers. Converted 5 POCs into paid pilots worth ~$40K within 60 days.
• Demo’d end-to-end MLOps pipeline: automated feature engineering → model selection → hyperparameter optimization.

Jun 2021 – Jan 2024 · 2 yrs 7 mos · Pittsburgh, PA · Remote

• Rebuilt fleet-safety risk model: preventable crash rate dropped 65% across 50 fleets, unlocking ~$3M in upsell revenue. Boosted AUROC from 0.64 to 0.81 with deep learning + automated HPO.
• Fine-tuned BERT with domain-specific embeddings on 500K+ documents for risk prediction across 10M+ miles monthly : saved customers $2.4M/yr in injury costs.

Aug 2020 – Jun 2021 · 10 mos · Remote

• Built the analytics function from scratch. Mined 120K exam responses to surface low-success questions, insights drove curriculum rewrite and raised pass rates 12%. Built sales forecasting (ARIMA, 10% MAPE) that cut $250K/yr in inventory costs.

Sep 2019 – May 2020 · 8 mos · Houston, Texas, United States · On-site

• Trained CNN to flag seismic noise vs. signal on 1M+ traces (F1 = 0.91), slashing manual QC effort time by 80 % for geophysicists
• Prototyped GCP REST service (~10 k preds/day demo) that proved feasibility of real-time QC and informed 2020 roadmap

Sep 2017 – May 2018 · 8 mos · Gainesville, Florida, United States

• Proposed the required propulsion motor after analyzing the mechanical and electrical requirement data for the electric pod
• Studied trade-offs between various motor topologies for high speed transportation systems
• Led vibration, shock, thermal, and functional simulation testing such as fluids, electrical, leak testing to determine that compliance and safety requirements are met
• Collaborated with the System Loads and Analysis team to design the wheel size and pre-load on the suspension system

Aug 2014 – May 2019 · 4 yrs 9 mos

• My work was in high efficiency thermoelectric materials. A technology that captures wasted heat from and converts it into useful electricity.
• Silicon nanowires with rough etched surfaces face an anomalous drop in thermal conductivity that's not explainable within the current theoretical framework.
• My PI and I invented a model to explain how that happens, and how this phenomenon can be leveraged to create high power, high efficiency thermoelectric generators.
• Thesis: ‘Non-linear non-equilibrium thermal transport in nanowires with rough surfaces’
• Collaborated with the research group from Comenius University in Slovakia and jointly discovered a universality in the structural parameters affecting thermal transport
• Expedited signal processing convolutions and achieved 82% reduction in computational cost by implementing memoization
• Presented research at the American Physical Society (APS) March Meeting in 2019, a conference attended by over 11,000 physicists over the globe
• Published 2 papers, one as a lead author, in leading Physics Journals:
• S. Abhinav and Muttalib, K.A. “Non-equilibrium phonon transport in surface-roughness dominated nanowires” (2019) J. Phys. Commun. 3 105010
• Muttalib, K. A., and S. Abhinav. "Suppressing phonon transport in nanowires: A simple model for phonon–surface-roughness interaction." (2017) Physical Review B 96.7 : 075403

Aug 2012 – May 2019 · 6 yrs 9 mos

• Prepared and delivered instruction to 3 sections of 18 students each, every semester for junior/senior undergraduate pre-med majors as well as engineering majors
• Deconstructed complex mathematical jargon into physical intuition of the concepts using animation, physical models and thought experiments that draw from day to day life
• Clearly communicated course syllabi, objectives, assignments, expectations for performance in the class including the tips to excel in the course via online access to the course webpage as well as 3-page physical handouts stating the same
• Delivered short videos in advance, via email, for students to understand the big idea behind the experiment
• Facilitated access to experimental setup during my office hours, a week in advance for students to tinker with
• Managed course related tasks such as recording student attendance and delivering final course grades as required by the program
• Collaborated with the course professors and other teaching assistants to create weekly quizzes and homework assignments
• Integrated collaborative learning by introducing group problem solving in the existing discussion sessions that drove attendance up from 50% to 90%
• Established an inclusive environment in the classroom by forming several small groups of 3-4 students each and allowing them to work together, improving the problem solving participation and reducing hesitation to ask for help
• Developed a strong credibility and an understanding of the diverse academic, cultural, socioeconomic and ethnic backgrounds of the students and faculty by relating to my own experiences as an international student